CS:Go On: Modeling Chained Loot Box Opening Behavior

Abstract

Video game loot boxes are virtual ``mystery boxes" that dispense random rewards when opened. Loot boxes have helped game developers monetize their games like never before, but legislators have sought to regulate the gaming mechanic due to its similarities to gambling. While there is extensive literature covering how to design loot boxes from a developer perspective and how to regulate them from a lawmaker perspective, research on how consumers themselves treat loot boxes is relatively sparse.

In this thesis, I will investigate consumer behavior towards loot boxes. In particular, I will pay special attention to the addiction-like phenomenon of chain opening, when consumers open several boxes in quick succession of each other. To construct a model for consumer chain-opening behavior, I first combine a large data set of CounterStrike: Global Offensive (CS:GO) openings with product data scraped from the game's secondary marketplaces. After assembling this data set, I utilize non-parametric and semi-parametric survival analysis techniques to model opening chains throughout their lifetimes and also quantify the importance of some of the main drivers of chain openings.